| Encoding: | UTF-8 |
| Type: | Package |
| Title: | K* Nearest Neighbors Algorithm |
| Version: | 0.1.2 |
| Description: | Prediction with k* nearest neighbor algorithm based on a publication by Anava and Levy (2016) <doi:10.48550/arXiv.1701.07266>. |
| License: | GPL-2 | GPL-3 [expanded from: GPL (≥ 2)] |
| Author: | Kei Nakagawa 
[aut, cre],
Shingo Sashida [aut] |
| Maintainer: | Kei Nakagawa <kei.nak.0315@gmail.com> |
| LazyData: | TRUE |
| Imports: | Rcpp |
| Depends: | R(≥ 3.0.2) |
| LinkingTo: | Rcpp(≥ 0.10.6) |
| RoxygenNote: | 6.1.1 |
| NeedsCompilation: | yes |
| Packaged: | 2019-03-28 09:53:44 UTC; SSunix |
| Repository: | CRAN |
| Date/Publication: | 2019-04-11 12:02:37 UTC |
This function calculates the prediction value of k* nearest neighbors algorithm.
Description
This function calculates the prediction value of k* nearest neighbors algorithm.
Usage
ksNN(Label, Distance, L_C = 1)
Arguments
Label |
vectors of the known labels of the samples. |
Distance |
vectors of the distance between the target sample we want to predict and the other samples. |
L_C |
parameter of k* nearest neighbors algorithm. |
Value
the prediction value(pred) and the weight of the samples(alpha).
Note
This algorithm is based on Anava and Levy(2017).
Examples
library(ksNN)
set.seed(1)
#make the nonlinear regression problem
X<-runif(100)
Y<-X^6-3*X^3+5*X^2+2
suffle<-order(rnorm(length(X)))
X<-X[suffle]
Y<-Y[suffle]
test_X<-X[1]
test_Y<-Y[1]
train_X<-X[-1]
train_Y<-Y[-1]
Label<-train_Y
Distance<-sqrt((test_X-train_X)^2)
pred_ksNN<-ksNN(Label,Distance,L_C=1)
#the predicted value with k*NN
pred_ksNN$pred
#the 'true' value
test_Y
This function calculates the prediction value of k* nearest neighbors algorithm.
Description
This function calculates the prediction value of k* nearest neighbors algorithm.
Usage
rcpp_ksNN(Label, Distance, L_C = 1)
Arguments
Label |
vectors of the known labels of the samples. |
Distance |
vectors of the distance between the target sample we want to predict and the other samples. |
L_C |
parameter of k* nearest neighbors algorithm. |
Value
the prediction value(pred) and the weight of the samples(alpha).
Note
This algorithm is based on Anava and Levy(2017).
Examples
library(ksNN)
set.seed(1)
#make the nonlinear regression problem
X<-runif(100)
Y<-X^6-3*X^3+5*X^2+2
suffle<-order(rnorm(length(X)))
X<-X[suffle]
Y<-Y[suffle]
test_X<-X[1]
test_Y<-Y[1]
train_X<-X[-1]
train_Y<-Y[-1]
Label<-train_Y
Distance<-sqrt((test_X-train_X)^2)
pred_ksNN<-rcpp_ksNN(Label,Distance,L_C=1)
#the predicted value with k*NN
pred_ksNN$pred
#the 'true' value
test_Y